Skip to Main content Skip to Navigation

Nano systèmes électromécaniques résonants à haute fréquence (NEMS HF) : une rupture technologique pour l'imagerie infrarouge non refroidi

Abstract : Progress in microelectronics has been mainly driven by informatics needs for addressing both increased performances and lower costs for processors and memories, according to the well-known Moore’s Law. For many years, these tremendous progresses in silicon fabrication and integration have also contributed to the emergence of new type of devices, such as sensors, actuators, filters, clocks or imagers, forming a new class of devices called More than Moore. Uncooled infrared imaging, which uses thermal sensors belongs to this new class of devices. Today thermal sensors principally use a thin resistive layer (mainly vanadium dioxide or amorphous silicon) on a suspended membrane as a thermometer and are called microbolometers. The fabrication cost of thermal cameras has dramatically dropped over the last 20 years, while attaining performances close to the expensive cooled cameras. Nevertheless, the cost of these imagers still remains too high for consumer market (night driving, smartphones, home automation) whereas military applications (surveillance, personal googles) need improved resolutions – in an affordable camera. Therefore, one objective of the microbolometers industry roadmap is to scale down the sensor surface – the pixel pitch – in order to increase the number of imagers fabricated on a silicon wafer. Yet, the pixel pitch reduction goes necessarily with a reduction of the captured infrared power leading to a reduction of the sensor signal. As a consequence, the sensor sensitivity needs to be improved as the pixel pitch scales down. The resistive technology has managed this scaling so far, down to 17µm pixel pitch, allowing a densification of the sensors by a factor 4 every 5 years. Despite this success, the scaling has been recently slowed down, mostly because of microbolometers self-heating issue and 1/f noise which are inherent to the resistive transduction. Our work has focused on a new type of sensor at 12µm pixel pitch, which theoretically gets rid of self-heating and 1/f noise. In our approach, an absorbing plate is excited at its mechanical resonance through two tiny torsion arms using an actuation electrode placed 2µm underneath. Pixel motion is also transduced electrostatically. Since micromechanical resonators feature very low frequency noise, we believe that an uncooled infrared sensor based on the monitoring of its resonance frequency (which changes with temperature through the TCF) should be extremely sensitive. In our work, we present different models (linear and nonlinear) for the pixel mechanical behavior and compare them to experimental characterization of resonators which were fabricated in dense arrays, according to several designs. We measure the frequency stability of our sensors along with their sensitivity to infrared flux. The best devices show a resolution of 30pW/sqrt(Hz), with a response time lower than one millisecond. The scene resolution (NETD) is 2K for an integration time compatible with imaging frame rate. These performances overtake results previously published on this topic with such reduced pixel pitch. We show that a NETD of 20mK (with a response time of 10ms) is reachable at 12µm pixel pitch if we can address the following 3 challenges: a cointegration of the resonators with their electronics, a shared readout of several pixels in the imaging frame rate and an improved TCF by a factor 10. Therefore, we provide different methods in order to improve the TCF. Finally, we present different pixel designs at 5µm pixel pitch which show theoretical performances close to uncooled infrared imaging requirements (NETD=70mK and tau_th=8ms). An optical transduction may also be a new route toward even better signal to noise ratio at low pitch.
Complete list of metadatas
Contributor : Abes Star :  Contact
Submitted on : Wednesday, March 14, 2018 - 11:16:08 AM
Last modification on : Friday, October 23, 2020 - 5:03:24 PM
Long-term archiving on: : Friday, June 15, 2018 - 2:12:30 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01731409, version 1




Ludovic Laurent. Nano systèmes électromécaniques résonants à haute fréquence (NEMS HF) : une rupture technologique pour l'imagerie infrarouge non refroidi. Optique [physics.optics]. Université Grenoble Alpes, 2017. Français. ⟨NNT : 2017GREAY050⟩. ⟨tel-01731409⟩



Record views


Files downloads